From: netmod <[email protected]> on behalf of Ladislav Lhotka <[email protected]> Sent: 09 June 2022 10:35
"Jan Lindblad \(jlindbla\)" <[email protected]> writes: > Hi NETMOD, > > A few days ago, as I was looking for examples of when-expressions in standard > modules, I stumbled over this in RFC 8944: > > augment "/nw:networks/nw:network" { > when '/nw:networks/nw:network/nw:network-types/l2t:l2-topology' { > description > "Augmentation parameters apply only for networks > with L2 topology."; > } > description > "Configuration parameters for the L2 network > as a whole."; > uses l2-topology-attributes; > } > > ... and a few other similar constructs in there. > > As you can see the when expression is unconstrained when it comes to which > network it refers to, which changes the semantics from "Augmentation > parameters apply only for networks with L2 topology" as the authors are > wishing for to "Augmentation parameters apply to all networks as soon as > there is at least one with L2 topology". > > Aside from RFC 8944, I also noticed very similar problems in RFC 9094, RFC > 8542, RFC 8294, RFC 8513 and RFC 8782. The list may be incomplete. Yes, I remember flagging this issue in at least one of my YD reviews. I think this also led to the myth that YANG Doctors prefer relative paths to absolute. :-) > > What do we do with the broken YANGs modules in these RFCs? A bis is needed in each case, errata unfortunately won't work. > > Since this seems to be a rather frequently occurring issue, are there any > mitigation steps we should take in NETMOD, apart from strengthening the YANG > Doctor review process? Hardly anything. It is IMO too late to part with XPath 1.0. On the one hand, it is a complicated beast, unnecessarily for YANG purposes. On the other hand, many YANG module authors don't bother to learn even the basics. > > What can we reasonably do to ensure all YANG Doctor reviews catch these > valid, but unintended XPath expressions before publication? Tools cannot help much with "false positive" results of XPath evaluation, we just have to be careful and double-check especially absolute paths. <tp> I have had similar issues when commenting on YANG modules such as 'augment' with no 'when' or the 'when' placed too low in the tree and the authors are often not interested. Their view seems to be that if the server supports the module, then it does not matter in how many places an irrelevant augment may happen. As long as the augment happens where it is needed, what is wrong with a few hundred or few thousand additional ones:-( Tom Petch Lada > > Best Regards, > /Jan Lindblad > > _______________________________________________ > netmod mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/netmod -- Ladislav Lhotka Head, CZ.NIC Labs PGP Key ID: 0xB8F92B08A9F76C67 _______________________________________________ netmod mailing list [email protected] https://www.ietf.org/mailman/listinfo/netmod _______________________________________________ netmod mailing list [email protected] https://www.ietf.org/mailman/listinfo/netmod
